airbourne sensing

Whether you’re interested in monitoring greenhouse gas emissions, locating airborne viruses, or just finding a mate by detecting individual pheromone molecules - as moths do - optical sensors, gas dispersion and statistical inversion are key fields. Provided you can measure trace concentrations well and understand how the atmosphere mixes as it moves, the answer to “who is emitting what, how much and where” - is literally Blowin in the wind. Read a technical article, and article in Significance and a recent blog in Medium about it.

ghost imaging

Can you retrieve a picture of an object through correlated measurements of a projected light using a single pixel camera and ghost imaging?

introductions to environmental extremes

With Jenny Wadsworth, I organised a workshop at Lancaster on extreme ocean wave phenomena in 2019. Jenny organised a prior meeting on interfaces in extreme value theory. Kevin Ewans and I contributed a chapter on extreme ocean conditions for the collection entitled "Ocean wave dynamics" edited by Ian Young and Alex Babanin of the University of Melbourne, which appeared in 2020. Matthew Jones and I recently wrote a blog in Medium on environmental extremes.

A dyma sgwrs YouTube ar eithafon amgylcheddol (a talk about environmental extremes in Welsh).

limit lines

Defining the "boundary" of a set of points in 2-D or higher dimensions is tricky. Many scientific disciplines face this problem. With Paul Carling (Southampton / Lancaster) and Teng Su (Chinese Academy of Sciences, Beijing), we wrote a review article seeking to outline a rational, structured approach to thinking about, defining and estimating these so-called "limit lines", with examples from environmental and geo-sciences.

model assurance

Data science is a huge field, and sometimes the empirical models on which it is based are not assessed well. This unpublished paper discusses some key issues in empirical model assessment.

molecular diffusion coefficients

The molecular diffusion coefficient quantifies the random thermal motion of molecules in a liquid or gas, depending on temperature, pressure, molecular and inter-molecular properties. They play an important role in many branches of chemical kinetics. With Peter Fisk (Shell), we developed new predictive models for diffusion coefficients of different solute / solvent combinations, from historical laboratory measurements, and wrote a book chapter about it.

multivariate spatial conditional extremes

With Rob Shooter (UK MetOffice) and Emma Ross (Shell), we examined extreme storms using satellite altimetry and scatterometry, and have provided software on GitHub for (multivariate) spatial conditional extremes analysis.

non-parametric extremes

Evandro Konzen and Claudia Neves at the University of Reading and I made a comparison of applied parametric (i.e. likelihood-based) and non-parametric inference for extremes.

non-stationary extremes

With Paul Northrop and David Randell, we've contributed a chapter on non-stationary extreme value modelling to the collection "Extreme value modeling and risk analysis" which appeared in book form in 2016.

pictish symbols and early medieval inscriptions

Ever wondered if ancient symbols have the characteristics of language? Here's a recent article and a follow-up study . It may also be possible to associate medieval inscriptions (like Pictish and Irish Ogam, Welsh Latin and Scandinavian Runes) with modern language lexicons. Read more here.

pragmatic engineering extremes

With Ed Mackay at the University of Exeter's Penryn Campus, we're seeking to provide straightforward engineering approaches incorporating recent developments in non-stationary and multivariate extremes.

With Ryota Wada and Takuji Waseda at the University of Tokyo, we're working on useful extreme value methods for small samples of tropical cyclones offshore Japan. Here are articles on the LWM (the likelihood weighted method) and STM-E (the spatio-temporal maximum - exposure model).

With Ross Towe and colleagues at Shell, we developed the covXtreme software for practical multivariate extreme value analysis with covariates, available here.

sparse gaussian processes

With Jeremy Sellier and Matthew Jones, supervised by Petros Dellaportas at University College London, we're are looking at point process models and reinforcement learning.

statistical planning and uncertainty analysis

With David Randell and Michael Goldstein at the University of Durham's School of Mathematical Sciences we developed optimal inspection methods for large industrial systems. The project applied Bayes linear methods to adjust beliefs about the integrity of large process systems. We wrote an article (here) for the Journal of Risk and Reliability illustrating the methodology in application to inspection design for offshore oil and gas facilities, and a second article on variance structure learning (here).

With Matthew Jones we applied the methodology to optimal design in remote sensing problems also. See here. We've recently written an article on optimal sequential design here, and are now working on Bayes linear analysis for ordinary differential equations, and higher-order Bayes linear.

return values

Here's a YouTube video on the whys and the why-nots of estimating return values subject to uncertainty.

satellite surrogate sensing

With Clay Roberts, supervised by Oli Shorttle and Kaisey Mandel at the Institute of Astronomy in Cambridge, Matthew Jones and I are looking at approaches to exploit satellite measurements of concentrations of surrogate species (e.g. NOx) to improve prediction of a target species (e.g. CH4) . Read more here.

wave impact

Predicting the forces generated by water waves hitting an angled wall is tricky, yet obviously important from an environmental engineering perspective. A team led by Alison Raby (Plymouth) measured wave impacts in a wave tank, and characterised the distributions of wave forces. Read more here.